strisunshinewentingwang/pytorch

Pull Requests Commits

Fix `check_compiler_ok_for_platform` on non-English locales (#85891)

Flamefire committed 1 year ago

fe87ae69

Re-introduce the functorch docs build (#85838)

zou3519 committed 1 year ago

0449cf0c

Pass group ranks and options to third party distributed backends (#73164)

esaliya committed 1 year ago

941d7a31

(bsr/csr) x dense mm (#85551)

nikitaved committed 1 year ago

e15a48de

Use `int64_t` for nll_loss with cuda inputs (#85395)

crcrpar committed 1 year ago

ef0baba2

resubmit: "resubmit: [mta] APEX style Fused Adam (#81705) (#85507)" (#85739)

crcrpar committed 1 year ago

5f26df03

Update debug flag for vulkan (#85715)

kimishpatel committed 1 year ago

44eefb13

Add vulkan qualifier to the kernel name (#85714)

kimishpatel committed 1 year ago

ad3bea58

[Pytorch][benchmark vulkan] Fix vulkan profiling (#85713)

kimishpatel committed 1 year ago

e0af12a0

Remove torch/extension.h dependency in torch/csrc/functorch/init.cpp (#85659)

zou3519 committed 1 year ago

e0170c7c

[fix] max_pool1d: shape check (#85594)

kshitij12345 committed 1 year ago

8fb470e8

catches failure on nvprim speculative lowering (#85580)

jjsjann123 committed 1 year ago

cab6ffa0

Stop cuda-10.2 binary builds (#85873)

atalman committed 1 year ago

a807f198

Add opt-einsum to CI (#85574)

janeyx99 committed 1 year ago

3cdf621f

Release GIL when doing shared memory copies on Tensors (#85389)

ezhang887 committed 1 year ago

b25a1ce2

Revert "C10D extension to enable per-thread PG (#84153)"

pytorchmergebot committed 1 year ago

6fae62b3

Separate magma installation for ROCm into its own file (#85567)

jithunnair-amd committed 1 year ago

976e2a35

Treat layout / pin_memory consistently across creation refs (#85333)

lezcano committed 1 year ago

9fb72ca4

Improve make_tensor performance for float and complex types (#85473)

peterbell10 committed 1 year ago

a76995e5

[qat]A more stable conv_bn fusion for qat training. (#85744)

Peizhao Zhang committed 1 year ago

ad87365e

[Profiler][trivial] Optimizer states (part 4 of Record Optimizer) (#85840)

slgong-fb committed 1 year ago

3cfc61b8

[c10d] move ncclgetlasterror directive definition upfront (#85825)

wanchaol committed 1 year ago

72b32f16

[ONNX] Update behavior for `register_custom_op_symbolic` (#85636)

justinchuby committed 1 year ago

dc63948d

Create a quantized non-in-palce version CUDA ReLU function, (#85669)

fufeisi committed 1 year ago

3c9e8cd8

[Profiler] bug fix: python object reference counting (#85847)

slgong-fb committed 1 year ago

7628603a

[ONNX] Fix reduce node shape inference (#85765)

BowenBao committed 1 year ago

edb99df2

Improve codegen for jvp decomposition (#84894)

soulitzer committed 1 year ago

7e468400

[Profiler] Use strong typedef for Tensor ID (#85718)

Taylor Robie committed 1 year ago

f23f362c

[Profiler] Fix traversal utility (#85717)

Taylor Robie committed 1 year ago

282d8dfa

[Profiler] Don't assign in AppendOnlyList::emplace_back (#85716)

Taylor Robie committed 1 year ago

dfdfaec3

Newer Older